Several recent conversations have touched on the quality of lidar data from southern Haiti collected for the World Bank shortly after the 12 January 2010 earthquake. In an effort to better understand these data, and to inform discussion about possible future lidar surveys in Haiti, I have looked more closely at the World Bank data using my CONSISTENCY analysis tool.
I have seen only limited documentation for these data. They were collected between 21-27 January 2010 with an airborne Leica ALS60 sensor. The aircraft was based in Santo Domingo, DR. Data collection and post-collection calibration appear to have been without any ground control in Haiti and without idifferential GPS base stations in or near the target area. Metadata documents hosted at OpenTopography reference TerraPOS PPP software, thus I infer that Precise Point Positioning was used to improve GPS solutions. The metadata documents indicate flying heights of 2500 - 6500 feet above ground level. I have not yet found any documentation of beam divergence, scan angle, or pulse repetition rate.
These data were collected and processed rapidly, with the primary
purpose of rectifying imagery collected on the same flights and making
DEMs and imagery available for post-earthquake relief efforts. The
metadata documents stress that flight plans were optimized for image
collection, not lidar collection, and that the processing (particularly
boresight calibration and ground point identification) were "ortho
grade".
These data are in the public domain. They may be downloaded from the IPLER website at the Rochester Institute of Technology.
These data are good!
P1, the estimated aggregate Z error, is ~4 cm. This is a respectably
low value typical of lidar surveys acquired under the best conditions
with extensive ground control. If this is typical of PPP GPS, the cost
of further lidar work in Hispaniola is likely to be less than I would
have guessed, as it appears that adequate data can be obtained without
close-by differential GPS base stations.
P2 value, the estimated aggregate XY error, is ~50 cm. This is not
so good. I'm accustomed to seeing values of 20-30 cm in similar
terrain. My guess is that the large XY error reflects the speedy
post-processing, and that with more time Kucera would have derived a
better pointing (boresight) calibration, though perhaps better pointing
calibration would require independent ground control.
With smaller XY errors, or processing data on a swath-by-swath basis
(so that swath mismatch is not a problem), I think a skilled TerraSCAN
operator could produce a better bare-earth DEM from these data.
Note on tile names: The LAS data are in files that correspond to 1 km x 1 km tiles, named according to the UTM coordinates of the lower-left corner. That is, file 76670435.las has data for the 1 km2 area with a lower left corner at 766,700m E, 2,045,500m N. To avoid impermissibly-long file names during GIS processing, I renamed tiles so that, for example, tile 76670435 has become x766y43.
| |
|
|
|
|
|
|
|
|
|
 x762y51 |
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
|
 x762y50 |
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
|
 x762y49 |
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
|
|
 x766y45 |
 x767y45 |
 x768y45 |
 x769y45 |
 x770y45 |
 x771y45 |
| |
|
|
|
|
|
|
|
|
|
 x762y44 |
 x766y44 |
 x767y44 |
 x768y44 |
 x769y44 |
 x770y44 |
 x771y44 |
| |
|
|
|
|
|
|
|
|
|
 x762y43 |
 x766y43 |
 x767y43 |
 x768y43 |
|
|
|
 x722y40 |
 x723y40 |
 x724y40 |
 x725y40 |
 x726y40 |
 x727y40 |
 x728y40 |
 x729y40 |
 x730y40 |
|
|
|
|
|
|
|
|
 x722y39 |
 x723y39 |
 x724y39 |
 x725y39 |
 x726y39 |
 x727y39 |
|
 x729y39 |
 x730y39 |
|
|
|
|
|
|
|
|
 x722y38 |
 x723y38 |
 x724y38 |
 x725y38 |
 x726y38 |
|
 x728y38 |
 x729y38 |
 x730y38 |
|
|
|
|
|
|
|
|
 x722y37 |
 x723y37 |
 x724y37 |
 x725y37 |
 x726y37 |
 x727y37 |
 x728y37 |
 x729y37 |
 x730y37 |
|
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y20 |
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y19 |
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y18 |
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y17 |
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y16 |
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y15 |
|
|
|
|
|
|
|
| |
|
|
|
|
|
|
|
|
 x761y14 |
|
|
|
|
|
|
|
| tile | RMSE | alpha95 | pulse/m2 | P1 - P2 (cm) |
| x768y44 | 2.989 | 0.162 | 4.41 | 6.2 33.2 |
| x761y14 | 0.342 | 0.673 | 0.86 | 27.9 898.1 |
| x762y44 | 0.186 | 0.391 | 4.98 | -4.5 51.6 |
| x761y15 | 0.174 | 0.393 | 2.48 | 16.7 0 |
| x768y43 | 0.167 | 0.113 | 6.63 | -31.3 0 |
| x767y44 | 0.150 | 0.335 | 4.11 | 4.8 50.4 |
| x767y45 | 0.141 | 0.270 | 2.81 | -10.8 17.6 |
| x762y43 | 0.125 | 0.275 | 4.26 | 2.8 37.8 |
| x766y43 | 0.124 | 0.236 | 3.09 | -0.5 35.6 |
| x771y45 | 0.120 | 0.252 | 4.57 | 3.3 28.1 |
| x766y44 | 0.120 | 0.262 | 3.69 | 6.5 54.2 |
| x727y40 | 0.111 | 0.206 | 1.32 | 8.6 0 |
| x769y45 | 0.106 | 0.192 | 4.26 | 0.7 39.7 |
| x770y44 | 0.094 | 0.215 | 5.52 | 2.9 41.2 |
| x729y40 | 0.094 | 0.152 | 1.94 | 9.9 40.3 |
| x770y45 | 0.093 | 0.203 | 4.49 | 1.2 32.5 |
| x728y40 | 0.090 | 0.178 | 1.83 | 8.2 28.3 |
| x768y45 | 0.089 | 0.186 | 3.81 | 1.8 22.4 |
| x729y37 | 0.086 | 0.153 | 3.07 | 8.1 32.1 |
| x730y40 | 0.084 | 0.009 | 1.09 | 9.4 104.2 |
| x725y37 | 0.080 | 0.165 | 3.16 | 4.0 36.1 |
| x769y44 | 0.078 | 0.149 | 5.82 | 3.8 26.0 |
| x730y39 | 0.075 | 0.146 | 3.9 | 6.0 30.5 |
| x728y37 | 0.075 | 0.010 | 3.02 | 4.9 34.8 |
| x762y50 | 0.071 | 0.139 | 4.68 | 2.3 34.1 |
| x724y39 | 0.071 | 0.142 | 1.34 | 6.0 0 |
| x767y43 | 0.069 | 0.159 | 3.59 | 48.8 0 |
| x722y39 | 0.067 | 0.129 | 1.04 | 5.9 18.3 |
| x723y39 | 0.066 | 0.129 | 1.09 | 5.6 79.5 |
| x722y38 | 0.066 | 0.127 | 1.03 | 5.4 30.8 |
| x762y51 | 0.065 | 0.123 | 4.12 | 4.3 34.2 |
| x771y44 | 0.064 | 0.134 | 5.88 | 5.3 28.4 |
| x762y49 | 0.064 | 0.129 | 4.85 | 5.6 29.3 |
| x728y38 | 0.061 | 0.116 | 4.33 | 3.1 37.7 |
| x723y38 | 0.059 | 0.114 | 1.33 | 5.1 26.2 |
| x727y37 | 0.058 | 0.108 | 3.32 | 4.0 33.9 |
| x727y39 | 0.054 | 0.098 | 3.74 | 4.5 39.9 |
| x730y38 | 0.053 | 0.097 | 4.08 | 3.2 34.8 |
| x724y37 | 0.053 | 0.102 | 3.08 | 3.7 40.9 |
| x725y39 | 0.052 | 0.102 | 3.2 | 4.1 26.7 |
| x729y39 | 0.051 | 0.095 | 3.96 | 3.7 38.2 |
| x726y37 | 0.051 | 0.098 | 3.45 | 2.7 29.6 |
| x724y38 | 0.050 | 0.100 | 3.8 | 4.7 30.9 |
| x723y37 | 0.046 | 0.088 | 3.21 | 3.3 33.6 |
| x761y19 | 0.044 | 0.081 | 3.86 | 2.9 16.4 |
| x726y38 | 0.044 | 0.085 | 4.4 | 3.4 29.3 |
| x730y37 | 0.043 | 0.083 | 3.27 | 2.9 28.1 |
| x726y39 | 0.042 | 0.083 | 3.55 | 2.6 35.8 |
| x722y37 | 0.042 | 0.079 | 2.66 | 4.0 53.8 |
| x761y17 | 0.040 | 0.079 | 3.79 | 2.2 18.9 |
| x729y38 | 0.040 | 0.077 | 4.22 | 1.6 38.7 |
| x761y16 | 0.039 | 0.077 | 3.79 | 2.5 20.9 |
| x761y18 | 0.037 | 0.077 | 3.85 | 2.1 17.0 |
| x725y38 | 0.036 | 0.069 | 4.43 | 3.1 30.7 |
| x761y20 | 0.032 | 0.064 | 3.16 | 2.3 16.9 |
| x766y45 | 0 | 0 | 1.83 | 0 0 |
| x726y40 | 0 | 0 | 1.41 | 0 0 |
| x725y40 | 0 | 0 | 1.07 | 0 0 |
| x724y40 | 0 | 0 | 0.97 | 0 0 |
| x723y40 | 0 | 0 | 0.78 | 0 0 |
| x722y40 | 0 | 0 | 0.66 | 0 0 |